ot3425662590
DESCRIPTION
International Journal of Engineering Research and Applications (IJERA) is an open access online peer reviewed international journal that publishes research and review articles in the fields of Computer Science, Neural Networks, Electrical Engineering, Software Engineering, Information Technology, Mechanical Engineering, Chemical Engineering, Plastic Engineering, Food Technology, Textile Engineering, Nano Technology & science, Power Electronics, Electronics & Communication Engineering, Computational mathematics, Image processing, Civil Engineering, Structural Engineering, Environmental Engineering, VLSI Testing & Low Power VLSI Design etc.TRANSCRIPT
E. K. Nyantakyi, C. A. Adams, J. K. Borkloe and D. Pobee / International Journal of Engineering
Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 4, Jul-Aug 2013, pp.2566-2590
2566 | P a g e
Synchronization of Signalised Intersections: A Case Study of Three
Major Intersections on the 24th February Road, Kumasi, Ghana
E. K. Nyantakyi1, C. A. Adams
2, J. K. Borkloe
3 and D. Pobee
4
1Department of Civil Engineering, Kumasi Polytechnic, P.O. Box 854, Kumasi, Ghana
2Department of Civil Engineering, KNUST, Kumasi, Ghana
3Department of Civil Engineering, Kumasi Polytechnic, P.O. Box 854, Kumasi, Ghana
4Department of Civil Engineering, Kumasi Polytechnic, P.O. Box 854, Kumasi, Ghana
ABSTRACT
Signalized intersections are critical elements
of an urban road transportation system and
maintaining these control systems at their optimal
performance for different demand conditions has
been the primary concern of the traffic engineers.
Traffic simulation models have been widely used in
both transportation operations and traffic analyses
because simulation is safer, less expensive, and
faster than field implementation and testing. This
study evaluated performance of selected signalized
intersections using micro simulation models in
Synchro/SimTraffic. Traffic, geometric and signal
control data including key parameters with the
greatest impact on the calibration process were
collected on the field. At 5% significant level, the
Chi square test and t-test analyses revealed that
headway had a strong correlation with saturation
flow compared to speed for both field and simulated
conditions. It was concluded that changes in phasing
plan with geometric improvement would improve
upon the selected signalized intersection’s level of
service. An interchange therefore should be
constructed at Anloga intersection to allow free
movement of vehicles thereby minimizing congestion
and accident occurrences. Stadium and Amakom
signalized intersections should be coordinated to
allow as many vehicles as possible to traverse those
intersections without any delay.
Key words: Performance measures, Sensitivity
Analysis, Signalized intersections, Synchro/ Sim
Traffic, S
I. INTRODUCTION In general, “simulation is defined as dynamic
representation of some part of the real world achieved
by building a computer model and moving it through
time” [1]. Computer models are widely used in traffic
and transportation system analysis. The use of computer
simulation started when D.L. Gerlough published his
dissertation: "Simulation of freeway traffic on a
general-purpose discrete variable computer" at the
University of California, Los Angeles, in 1955 [2].
From those times, computer simulation has become a
widely used tool in transportation engineering with a
variety of applications from scientific research to
planning, training and demonstration.
The 24th February Road is an East-West
principal arterial of about 5.4 km length from KNUST
junction to the UTC traffic light. The road is a 2-lane
dual carriageway, and paved over its entire length. The
road provides the main route that leads into the Kumasi
metropolis from the southern parts of Ghana. The
Anloga, Stadium and Amakom signalized intersections
are three major intersections on one of the major
arterials of Kumasi. These selected signalized
intersection approaches are traversed by the same main
arterial road entering Kumasi from Accra. They share
similar traffic and driver characteristics. The road
corridor is relatively heavily trafficked [3]. The
intersections are characterized by long vehicular queues
at the approaches of the intersections, especially during
morning and evening peak periods of the day. However,
for some time now observation of the current traffic
situation at the selected intersections shows that very
long queues of vehicles form on the intersection legs.
Traffic congestion levels continue to rise daily at the
selected intersections and there are significant travel
time delays and lower levels of service [3].
Previous studies by [3] on the performance of the
intersections attributed congestion to critical capacity,
intersection controls and abuse to motorists and/or
pedestrians. It was concluded that most of the sections
on the 24th
February road have their capacities
approaching critical (v/c ratio>0.6) and that this
contributes in part to the congestion which results in
delay and subsequent poor performance of the road. It
was therefore recommended that this section of the 24th
February road will be highly critical in the next 3 to 5
years (period from 2004 – 2009). As part of the
recommendations [3], the report proposed to improve
upon the signalization and capacity at the intersection
through; revision of signal timing, phasing plan, lane
assignment/designation and inclusion of exclusive
NMT phase phase at Anloga Junction, revision of
signal timing and phasing plan and inclusion of
E. K. Nyantakyi, C. A. Adams, J. K. Borkloe and D. Pobee / International Journal of Engineering
Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 4, Jul-Aug 2013, pp.2566-2590
2567 | P a g e
concurrent NMT phase at Stadium intersection, revision
of signal timing, phasing plan and inclusion of
exclusive NMT phase and removal of illegal taxi rank
at the filling station on the Asawasi leg at Amakom
intersection.
In the study although the micro simulation
Synchro/SimTraffic was reportedly used for the
analysis however, the models were not calibrated. Even
though some of the recommendations have been
implemented at the selected intersections, long queues
and frequent delays still persist during peak hour
conditions [3]. Since in the application of micro
simulation tools, one major step is calibration or
adaptation before the prediction can be said to mimic
site conditions, this needs to be checked. Also micro
simulation tools application is relatively new in Ghana
and the procedures for calibration, and application in
modeling is not very well understood by practising
engineers. Outputs of micro simulation models are
useful to traffic engineers to identify existing problems
and come out with interventions. A micro simulation
model in Synchro is a software that makes use of
limited data by trying to mimic the present traffic
situation. This is then used in forecasting the future
traffic conditions based on the present. This study seeks
to contribute to the knowledge base in the area by
calibrating the Synchro/SimTraffic models and using
the concept to predict the performance of the selected
intersections. This study therefore evaluated
performance of the selected signalized intersections
using micro simulation models in Synchro/SimTraffic.
II. METHODOLOGY 2.1 Calibration Procedure
The calibration procedure employed [4]
approach by first selecting the intersection, collecting
required input data for the Synchro model and
collecting calibration data. These were then followed by
comparing calibration data with simulated results from
the field and finally calibrating the model.
2.2 Site Selection and Description
The selected signalized intersections were
selected based on their accident and safety records in
the past and also the levels of congestion associated
with these intersections. The intersections were also
selected for easy coordination. Fig. 1 shows the map of
Kumasi showing the selected intersections.
Figure 1: Map of Kumasi showing the selected signalized intersections on the 24th
February Road
Source: from Department of Urban Roads, Kumasi-
Ghana
2.2.1 Anloga Intersection
The Anloga junction is a signalised
intersection comprising four principal arterials.
It is about 2.6 km West of the KNUST junction. The
intersection is 600m away from Oforikrom traffic light
and 950m away from the Stadium traffic light. The
intersection has four legs. Fig. 2 shows the geometry of
Anloga intersection.
E. K. Nyantakyi, C. A. Adams, J. K. Borkloe and D. Pobee / International Journal of Engineering
Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 4, Jul-Aug 2013, pp.2566-2590
2568 | P a g e
Figure 2: Geometry of Anloga Intersection
Source: from study
2.2.2 Stadium Intersection
The intersection with Hudson road is
signalized and about 3.6km west of the KNUST
junction. It is 950m away from Anloga junction and
350m away from the Amakom traffic light. The
intersection has three legs with one approach/entry and
exit lanes on the minor road, (Hudson road), and two
approach/entry and exit lanes on the 24th February road.
The two approach/entry and exit lanes road is a 2-lane
dual carriageway, which is paved over its entire length.
It is the intersection of a Principal arterial and a Minor
arterial. Figure 3 is sectional map of Kumasi showing
the Stadium intersection. Fig. 3 shows the geometry of
stadium intersection.
Figure 3: Geometry of 24th
February/Hudson Road Intersection
Source: from study
2.2.3 Amakom Intersection
The Amakom traffic light, formerly called the
Amakom roundabout, is a signalised intersection. It is
about 4km West of the KNUST junction. The
intersection is 350m away from Stadium traffic light
and 700m away from labour roundabout. The
intersection has four legs with one approach/entry and
exit through lanes on each leg of the minor road, (Yaa
Asantewaa road), and two approach/entry and exit
through lanes on the 24th February road. It is the
intersection of a Principal arterial and a Collector road.
E. K. Nyantakyi, C. A. Adams, J. K. Borkloe and D. Pobee / International Journal of Engineering
Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 4, Jul-Aug 2013, pp.2566-2590
2569 | P a g e
Fig. 4 is the map of Kumasi showing the intersection
under study.
Figure 4: Geometry of 24th
February/ Yaa Asantewaa Road Intersection
Source: from study
2.3Basic Theoretical Background
The concept of “Car-following” describes the
detailed movement of vehicles proceeding close
together in a single lane. This theory is based on the
assumption that each driver reacts in some specific
fashion to a stimulus from the vehicle ahead of him.
One of the oldest and most well known cases of the use
of simulation in theoretical research is the “car-
following” analysis based on the Generalized General
Motors (GM) models. In these models a differential
equation governs the movement of each vehicle in the
platoon under analysis [5]. Car-following, like the
intersection analysis, is one of the basic equations of
traffic flow theory and simulation, and the analysis has
been active after almost 40 years from the first trials [6].
The car-following theory is of significance in
microscopic traffic flow theory and has been widely
applied in traffic safety analysis and traffic simulation
([7]; [8]). There have been many car-following models
in the past 60 years, and the models can be divided into
two categories. One is developed from the viewpoint of
traffic engineering and the other is based on statistical
physics. From the perspective of traffic engineers [9],
car-following models can be classified as stimulus-
response models ([10]; [11]), safety distance models
[12], psycho-physical models [13], and artificial
intelligence models ([14]; [15]).
The car-following theory is based on a key assumption
that vehicles will travel in the center line of a lane,
which is unrealistic, especially in developing countries.
In these countries, poor road conditions, irregular
driving discipline, unclear road markings, and different
lane widths typically lead to non-lane-based car-
following driving [16]. Heterogeneous traffic,
characterized by diverse vehicles, changing
composition, lack of lane discipline, etc., results in a
very complex behavior and a non-lane-based driving in
most Asian countries [17]. Therefore, it is difficult for
every vehicle to be moving in the middle of the lane.
Vehicles are positioned laterally within their lanes, and
the off central-line effect results in lateral separations.
However, to the limit of our knowledge, the effect of
lateral separation in the car-following process has been
ignored by the vast majority of models. A few
researchers have contributed efforts on this matter. [16]
first developed a car-following model with lateral
discomfort. He improved a stopping distance based
approach that was proposed by [12], and presented a
new car-following model, taking into account lateral
friction between vehicles.
[18] proposed a non-lane-based car following model
using a modified full-velocity difference model. All the
above models have assumed that drivers are able to
perceive distances, speeds, and accelerations. However,
car-following behavior is a human process. It is
difficult for a driver of the following vehicle to perceive
minor lateral separation distances, and drivers may not
E. K. Nyantakyi, C. A. Adams, J. K. Borkloe and D. Pobee / International Journal of Engineering
Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 4, Jul-Aug 2013, pp.2566-2590
2570 | P a g e
have precise perception of speeds and distances, not to
mention accelerations.
2.3.1 Car-following Models
The logic used to determine when and how
much a car accelerates or decelerates is crucial to the
accuracy of a microscopic simulation model. Most
simulation models use variations on the GM model.
Although it was developed in the 1950s and 1960s, it
has remained the industry standard for describing car-
following behavior and continues to be verified by
empirical data. A variation on the GM model is the
PITT car-following model, which is utilized in
FRESIM. The GM family of models is perceived to be
the most commonly used in microscopic traffic
simulation models and are, therefore, the focus of this
article.
2.3.1.1Generalized General Motors Models
The first GM model modeled car-following is
a stimulus-response process in which the following
vehicle attempts to maintain space headway. When the
speed of a leading vehicle decreases relative to the
following vehicle, the following vehicle reacts by
decelerating. Conversely, the following vehicle
accelerates when the relative speed of the leading
vehicle increases. This process can be represented by
the first GM model, given below:
tt FLF F Eq. (1)
Where: F
= acceleration of the following
vehicle,
F
= speed of the following
vehicle,
L
= speed of the leading vehicle,
αF = sensitivity of the following
vehicle, and
t = time.
2.3.1.2 PITT Car-following Model
FRESIM uses the PITT car-following model,
which is expressed in terms of desired space headway,
shown in the equation below.
2212 TVtVbkkVmLths Eq. (2)
Where: hs(t) = desired space headway at time t,
L = length of leading vehicle,
m = minimum car-following distance
(PITT constant),
k = car-following sensitivity factor
for following vehicle,
b = relative sensitivity constant,
v1(t) = speed of leading vehicle at
time t, and
v2(t) = speed of following vehicle at
time t.
Equation above can be solved for the following
vehicle’s acceleration, given by the equation below.
KTT
tVtVbkTKVmLyxa
2
22
2
212
Eq. (3)
Where: a = the acceleration of the following vehicle,
T = the duration of the scanning
interval,
x = position of the leading vehicle,
and
y = position of the following vehicle.
2.4Algorithm on Synchro/SimTraffic software
Simulation is basically a dynamic
representation of some part of the real world achieved
by building a computer model and moving it through
time. The results obtained from any simulation model
will be as good as the model replicates the specific real
world characteristics of interest to the analyst.
Once a vehicle is assigned performance and driver
characteristics, its movement through the network is
determined by three primary algorithms:
2.4.1 Car following
This algorithm determines behavior and
distribution of vehicles in traffic stream. Synchro varies
headway with driver type, speed and link geometry
whereas SimTraffic generates lower saturation flow
rates.
2.4.2 Lane changing
This is always one of the most temperamental
features of simulation models. There are three types of
lane-changing which includes
Mandatory lane changes (e.g., a lane is obstructed
or ends)
Discretionary lane changes (e.g., passing)
Positioning lane changes (e.g., putting themselves
in the correct lane in order to make a turn): There is
heavy queuing and this is a common problem for
modeling positioning lane changes. Vehicles often
passed back of queue before attempting lane
E. K. Nyantakyi, C. A. Adams, J. K. Borkloe and D. Pobee / International Journal of Engineering
Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 4, Jul-Aug 2013, pp.2566-2590
2571 | P a g e
change and their accuracy relates to degree of
saturation and number of access points such as
congested conditions which requires farther look
ahead and densely-spaced access (i.e. short
segments) which presents a problem.
2.4.3 Gap Acceptance
Gap acceptance affects driver behavior at
unsignalized intersections, driveways (e.g., right-in-
right-out) and right-turn-on-red (RTOR) movements. If
default parameters are too aggressive, vehicle delay will
be underestimated and there is serious implication for
frontage roads. Conversely, parameters which are too
conservative may indicate need for a signal when one
isn’t necessary. Gap acceptance parameters are
network-wide in SimTraffic.
2.5 Data Collection for Synchro
Microscopic simulation model Synchro has
many model parameters. In order to build a Synchro
simulation model for the selected intersections and to
calibrate it for the local traffic conditions, two types of
data are required. The first type is the basic input data
which include data on network geometry, traffic
volume, turning movements and traffic control systems.
The second type is the observation data employed for
the calibration of simulation model parameters such as
average link speeds, headways and saturation flow
using standard procedures.
2.5.1 Videotaping Traffic at Anloga Intersection
Anloga intersection was filmed because of the
possibility of obtaining good elevation observer
positions as well as capturing the complex traffic
situations that exist at the intersection. The Anloga
intersection was filmed for three (3) hours, from
0700hours and 1000 hours on Wednesday, the 22nd
of
April 2012. A total of 50 cycles during the morning
peak conditions were captured. Two digital cameras
were mounted on a tripod stand on top of a building at
Anloga intersection to record traffic volumes and
queues, headways and signal heads. The digital video
recorder has the capability to record and playback
simultaneously the signal from the cameras. This allows
for relating in real-time discharging traffic with the
state of the traffic signals. The video images on DVD
were played back and analyzed manually on a laptop
computer by number of trained observers. Traffic
volumes and headways were collected from the films.
Actual site observations and traffic signal timings were
determined manually using stop watches at the
intersection and checked with the designer’s data. The
films were played back and the site conditions also
observed to explain some of the values obtained.
2.5.2 Extracting Data from Tapes
In the third phase of data collection,
videotapes were played back and traffic volumes and
headway data were extracted. First, the vehicles were
carefully counted and time-measured while playing
back the tapes. The data extraction was conducted on a
lane-by-lane basis. Then, these measurements were
used to calculate intermediate results and estimate
saturation flows. Table 1 shows the traffic volume data
and geometric data for Anloga intersection.
2.5.3 Manual Collection of Data at Stadium and
Amakom Intersections
Manual collection of data was done at Stadium
and Amakom intersections because it was difficult in
getting good elevation observer positions as well as the
fact that traffic situations are not as complex as Anloga
intersection.
Traffic volume data and geometric data as
indicated in Tables 2 and 3 for turning movements were
collected manually at Stadium and Amakom
intersections on Wednesday, the 13th
and 27th
of May
2012 respectively between 0700hours and 1000 hours
during the morning peak period of the day. Traffic
signal timings were also determined manually using
stop watches at the Stadium and Amakom intersection.
At Stadium and Amakom intersections, two
enumerators each were positioned on each approach.
The number of vehicles turning left, right and through
traffic were counted and the times when an approach
has green and red indications were recorded. Two other
enumerators each also recorded headways and speeds
of vehicles as they traverse the intersections.
From Tables 1, 2 and 3, it can be observed that
a peak hour volumes of 3,660, 2,434 and 2,980 vehicles
respectively were registered at Anloga, Stadium and
Amakom intersections respectively. These values
represented the worst traffic situation for an average
day. Therefore, the existing control needs to be replaced
with a more effective, efficient and reliable control
scheme to ensure smooth and safe operations of all
types.
E. K. Nyantakyi, C. A. Adams, J. K. Borkloe and D. Pobee / International Journal of Engineering Research and Applications (IJERA)
ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 4, Jul-Aug 2013, pp.2566-2590
2572 | P a g e
Table 1: Summary Peak Hour Traffic Volume and Geometric Data at Anloga Intersection
% No Lane Storage % App Vol % of total Total Int
Movement
Heavy of Width Lengths
of
App (veh/hr) Int Vol
From To Code Veh/hr Veh Lanes
Vol
volume (veh/hr)
Anloga WBL 7 50 1 3.0 75 0.5
KNUST Amakom WBT 1013 3 2 3.4
76.1 1332 36.4
Airport R/A WBR 317 7 1 3.0 75 23.4
Airport R/A EBL 291 12 1 3.4 86 18.8
Amakom KNUST EBT 1211 6 2 3.3
78.4 1545 42.2
Anloga EBR 43 15 1 4.7 54 2.8
3660
Amakom NEBL 141 9 1 4.8
63.5
Anloga Airport R/A NEBT 30 40 1 4.8
13.5 222 6.1
KNUST NEBR 51 21 1 4.7 34 23.0
KNUST SWBL 144 16 1 3.0 65 25.7
Airport R/A Anloga SWBT 20 60 1 3.8
3.6 561 15.3
Amakom SWBR 3974 10 1 3.8 58 70.7
Source: from study
E. K. Nyantakyi, C. A. Adams, J. K. Borkloe and D. Pobee / International Journal of Engineering Research and Applications (IJERA)
ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 4, Jul-Aug 2013, pp.2566-2590
2573 | P a g e
Table 2: Summary Peak Hour Traffic Volume and Geometric Data at Stadium Junction Intersection
% No Lane Storage % App Vol
% of
total Total Int
Movement
Heavy of Width Lengths
of
App (veh/hr) Int Vol
From To Code Veh/hr Veh Lanes
Vol
volume (veh/hr)
Stadium Jn WBL 432 8 1 2.9 73 30.9
Anloga Jn Amakom WBT 967 1 2 3.5
69.1 1399 57.5
N/A WBR
N/A EBL
Amakom Anloga Jn EBT 805 5 2 3.7
95.2 846 34.8 2434
Stadium Jn EBR 41 13
shared 4.8
Amakom NBL 106 4 1 3.6
56.1
Stadium N/A NBT
189 7.7
Anloga Jn NBR 83 20 1 2.7 34 43.9
Source: from study
Table 3: Summary Peak Hour Traffic Volume and Geometric Data at Amakom Intersection
% No Lane Storage % App Vol % of total Total Int
Movement
Heavy of Width Lengths
of
App (veh/hr) Int Vol
From To Code Veh/hr Veh Lanes
Vol
volume (veh/hr)
Amakom WBL 117 22 1 3.8 56 8.5
Anloga Jn Kejetia WBT 1153 3 2 2.9
83.4 1383 46.4
Asawasi WBR 113 24 1 4.5 43 8.1
Asawasi EBL 76 7 1 3.6 46 8.6
Kejetia Anloga Jn EBT 742 2 2 3.6
83.6 886 29.7
Amakom EBR 69 0
54 7.8
2980
Kejetia NBL 53 9
shared 24.7
Amakom Asawasi NBT 153 17 1 3.4
71.2 215 7.2
Anloga Jn NBR 9 100
16 4.1
Anloga Jn SBL 120 13 1 4 23 24.2
Asawasi Amakom SBT 222 13 1 4
44.8 496 16.7
Kejetia SBR 154 7 1 3 22 31
Source: from study
E. K. Nyantakyi, C. A. Adams, J. K. Borkloe and D. Pobee / International Journal of Engineering
Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 4, Jul-Aug 2013, pp.2566-2590
2574 | P a g e
2.5.4 Signal Control Data for Selected Intersections
Cycle length is the time required for one
complete sequence of signal indications (phases).
Usually it is measured in seconds. Cycle lengths and
phases for the intersection were recorded using
stopwatch. Cycle lengths for the selected signalized
intersection were obtained as 210, 68 and 172 seconds
respectively. Table 4 shows the signal timing data for
the selected intersections.
Table 4: Signal Timing Data for Selected Intersections
Intersection Location Cycle
Length
(C)
Actual
Green
Time (G)
Actual
Yellow
Time (Y)
Actual Red
Time (R)
Total Lost
Time
(l1+l2)
Effective
Green
Time (g)
Anloga From KNUST 210 93 4 2 4 95
To KNUST 210 93 4 2 4 95
Stadium From KNUST 68 35 4 2 4 37
To KNUST 68 30 4 2 4 32
Amakom From KNUST 172 56 4 2 4 58
To KNUST 172 47 4 2 4 49
Source: from study
The resulting effective green time (g) was therefore
calculated using the equation below
)21
( llRYGg Eq. (4)
Where l1 = start up lost time
l2 = clearance lost time
2.6 Calibration Data for Selected Intersections
In order to realistically model traffic at the
intersection, it is important to have realistic calibration
data. To calibrate the model for local condition speeds
and Headways data were collected.
2.6.1 Spot Speed Data
Speed is the most important parameter
describing the state of a given traffic stream. In a
moving traffic stream, each vehicle travels at a different
speed. Thus, the traffic stream does not have a single
characteristic speed but rather a distribution of
individual vehicle speeds. The spot speed data to and
from KNUST approaches at the selected intersections
were collected using the Doppler principle (radar). The
speed data were collected as the tail of the vehicles
crosses the stop bar. The first four vehicles in queue
were not counted because they were accelerating from
rest. A radar gun was operated by a single person.
Operators randomly targeted the vehicles and recorded
the digital readings displayed on the unit. Through
traffic speeds to and from KNUST approaches at
Anloga intersection were recorded for a maximum of
60 vehicles inclusive of private cars, commercial
vehicles and heavy goods vehicles. Similarly, through
traffic speeds from and to Tech approaches at Stadium
and Amakom intersections were recorded for a
maximum of 30 vehicles and 60 vehicles inclusive of
private cars, commercial vehicles and heavy goods
vehicles. The unit of speed is in km/hr.
2.6.2Headway and Saturation Flow Data
The headway is the time starting from when
the tail of the lead vehicle crosses the stop bar until the
front of the following car crosses the stop bar. Headway
data for three cycles each were collected at the
intersection using stop watches. Tables 5, 6 and 7 show
the summary of computed saturation flow, headway and
speed for the selected intersections. The unit of
saturation flow is passenger car units per hour per lane
(pcu/hr/lane) and that of headway is in seconds (sec).
2.7 Calibration of the Synchro Models
The successful utilization of the model
depends on selecting the proper values of the
parameters that describe the traffic performance and
driving behavior characteristics. Synchro/SimTraffic
model is carefully calibrated and validated to provide
meaningful results.
2.7.1 Chi Square Test Analysis (P-value)
This was used to determine the level of
significance between the computed and simulated
saturation flows, speeds and headways at the selected
intersections. When P< 0.05, it was considered
significant and P> 0.05 was considered not significant.
2.7.2 Paired sample T-Test Analysis
This analysis was carried out to either confirm
or reject the chi squared test analysis. It was also used
to evaluate the variation in the computed and simulated
saturation flows, speeds and headways at the selected
intersections.
E. K. Nyantakyi, C. A. Adams, J. K. Borkloe and D. Pobee / International Journal of Engineering
Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 4, Jul-Aug 2013, pp.2566-2590
2575 | P a g e
2.7.3 Regression Analysis
Calibration of the model was based on
regression analysis using traffic volume data to-and-
from KNUST approaches at the selected intersections.
It was carried out to establish whether speed or
headway had a strong correlation with saturation flow.
The predictors were speed and headway whiles the
dependent variable was saturation flow.
2.8 Performance Assessment for the Selected
Intersections
Levels of service and delay were used to
evaluate the performance at the selected intersections.
A Level of service is a letter designation that describes
a range of operating conditions on a particular type of
facility. The 1994 Highway Capacity defines levels of
service as “qualitative measures that characterize
operational conditions within a traffic stream and their
perception by motorists and passengers.” Six levels of
service are defined for capacity analysis. They are
given letter designations A through F, with LOS A
representing the best range of operating conditions and
LOS F the worst. Delay is defined in terms of the
average stopped time per vehicle traversing the
intersections.
2.8.1Change of Phase without Geometric
Improvement
For effective investigation of the optimized
cycle lengths and offsets at the selected signalized
intersections, three different alternatives with four-
phase operational plan in were proposed at Anloga.
Similarly, three (3) different alternatives with three-
phase operational plan were proposed at Stadium
intersection. Furthermore, two (2) different alternatives
with four-phase operational plan were proposed at
Amakom intersection.
2.8.2 Change of Phase with Geometric Improvement
The best alternative chosen was further
investigated upon by the addition of lane(s) to the
through put traffic at the selected intersections. This
was done to determine whether there was improvement
in the level of service. This is because improvement in
the Level of Service (LOS) at the selected intersections
would result in overall and enhanced performance on
the 24th
February road corridor.
2.8.3 Grade Separation Option
Continuous addition of lanes to the through put traffic
will adversely affect the reservations at the intersection
and when that condition persists, grade separation
option will be considered to see how best to allow free
flow of traffic to traverse the intersection concerned.
2.9 Sensitivity Analysis
This was carried out to verify in detail the
sensitivity of the obtained results to the variation of the
input parameters at the selected intersections.
III. RESULTS AND DISCUSSION 3.1 Saturation flow, Headway and Speed Data
Table 5 shows the summary of computed saturation
flows and headways for the selected intersections.
Table 5: Computed saturation flow, headway and speed
Intersection Performance
measures
Direction No. of
vehicles
Mean Maximum Minimum Standard
Deviation
Saturation flow From KNUST 60 1662 4186 467 733.89
To KNUST 60 1742 3913 403 712.85
Anloga Headway From KNUST 60 2.59 7.71 0.86 1.29
To KNUST 60 2.56 8.93 0.92 1.51
Speed From KNUST 60 26.63 36 20 3.89
To KNUST 60 28.43 46 20 6.11
Saturation flow From KNUST 30 1226 2707 360 570.71
To KNUST 30 1643 2727 678 511.65
Stadium Headway From KNUST 30 3.7 10.0 1.33 1.99
To KNUST 30 2.44 5.31 1.32 0.89
Speed From KNUST 30 26.9 33 22 3.02
To KNUST 30 24.8 28 22 1.67
Saturation flow From KNUST 60 1489 3214 497 726.58
Amakom To KNUST 60 1286 3186 256 582.51
Headway From KNUST 60 3.01 7.24 1.12 1.42
To KNUST 60 3.54 14.08 1.13 2.18
Speed From KNUST 60 26.3 32 23 2.58
To KNUST 60 26.5 30 22 1.96
Source: from study
E. K. Nyantakyi, C. A. Adams, J. K. Borkloe and D. Pobee / International Journal of Engineering
Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 4, Jul-Aug 2013, pp.2566-2590
2576 | P a g e
3.1.1 Explanation of Saturation Flow Values
Saturation flow, which is the maximum rate of
flow of traffic across the stop line at an intersection, is a
very important measure in junction design and signal
control applications. Low values of Saturation flow
means less vehicles can cross the stop line when the
signal turns green. Data collection techniques for the
determination of saturation flow are well elaborated by
[19].
The saturation flow from KNUST junction
approach was 1662pcu per hour per lane and that to
KNUST junction approach was 1742pcu per hour per
lane at Anloga intersection.
Also, for the stadium intersection, the saturation flow
rates from Tech approach was 1226pcu per hour per
lane and that to Tech approach was 1643pcu per hour
per lane. Similarly, the saturation flow rates from Tech
approach was 1489pcu per hour per lane and that to
Tech approach was 1254pcu per hour per lane for the
Amakom intersection.
Video playbacks for Anloga intersection
allowed for the observation of extraneous factors
contributing to the reduction in saturation flows and
flow interruptions at approaches for specific times and
cycles some of which were rejected for analysis due to
flow interruptions. This was attributed to vehicle mix,
geometry of intersection, driver behaviour, public
transport proportion in traffic stream, stops near
intersection along routes (within 20m) and pedestrian
indiscriminate crossing due to location of attractions
and roadside activity. These were identified as principal
factors affecting flow, which collaborates [20]; [21];
and [22], who have severally reported similar results.
The presence of fuel service stations, high pedestrian
volumes and erratic pedestrian crossing and roadside
activities significantly affected the saturation flow rates
at the Anloga intersection. High approach down grades
and the presence of heavy vehicles at Anloga
intersection were also found to be major contributory
factors to the deviation of the saturation flow from the
ideal.
3.2Results of Calibration of Synchro Models for
Selected Intersections
3.2.1 Chi-squared Analysis (p-values)
Table 6 shows the results of the comparison
between computed and simulated saturation flows,
speeds and headways for the selected intersections
using the Chi Square Test (p-value).
Table 6: Comparison of Computed and Simulated Performance measures
Intersection Performance
measures
Direction Computed
values Sc
Simulated
values, Sa
Ratio
(Sc/Sa)
Chi square
Test (P-
values
Saturation flow From KNUST 1662 1700 0.978 0.3269
To KNUST 1742 1756 0.992
Anloga Headway From KNUST 2.59 2.97 0.872 0.6525
To KNUST 2.56 3.27 0.783
Speed From KNUST 26.63 23 1.103 0.2383
To KNUST 28.43 24 1.185
Saturation flow From KNUST 1226 1267 0.968 0.2494
To KNUST 1643 1643 1.000
Stadium Headway From KNUST 3.7 3.06 1.209 0.58011
To KNUST 2.44 3.18 0.767
Speed From KNUST 26.9 25 1.076 0.4792
To KNUST 24.8 22 1.127
Saturation flow From KNUST 1489 1535 0.970 0.2398
To KNUST 1286 1288 0.998
Amakom Headway From KNUST 3.01 3.36 0.896 0.6902
To KNUST 3.54 2.94 1.204
Speed From KNUST 26.3 21 1.252 0.1714
To KNUST 26.5 23 1.152
Source: from study
P-values > 0.05 meant that there were no
significant differences between the computed and
simulated values in terms of saturation flow, headway
and speed data for the selected signalized intersections.
Table 6 indicated that the field saturated flow, headway
and speed values were similar to the simulated saturated
flow, headway and speed values obtained from Synchro
for the selected intersections.
E. K. Nyantakyi, C. A. Adams, J. K. Borkloe and D. Pobee / International Journal of Engineering
Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 4, Jul-Aug 2013, pp.2566-2590
2577 | P a g e
3.2.2 T-test analysis
For saturation flow at each approach of the
selected intersections, the test results showed that there
was no significant (p>0.05) difference between the
computed and simulated saturation flow values. For
headways at each approach of the selected intersections,
the test results showed that there was no significant
(p>0.05) difference between the computed and
simulated headway values.
For speeds however at each approach of the
selected intersections, the test results showed that there
was significant (p<0.05) difference between the
computed and simulated speed values. It was found that
a certain percentage of the variation in the simulated
speed values was explained by the field speed values
and Eta squared values explained in Tables 7, 8 and 9.
Detailed Results of level of significance test carried out
at the selected Intersections (T-Test Analysis)
Table 7: Results of paired Sample Test at Anloga Intersection (Sat flow SIM)
Approaches
Paired Difference t Eta
squared
(%)
Sig.
(2-tailed)
p-value Mean Standard
deviation
Standard
error
mean
95% confidence
interval of the
difference
Lower Upper
Saturation
from KNUST
-38.033 729.842 94.222 -226.6 150.505 -0.404 0.28 0.688
Saturation to
KNUST
-14.450 711.929 91.910 -198.361 169.461 -0.157 4.2 0.876
Speed from
KNUST
-3.650 1.696 0.219 -4.088 -3.212 -16.673 82.5 0.000
Speed to
KNUST
4.433 5.444 0.703 3.027 5.840 6.308 4.2 0.000
Headway from
KNUST
-0.673 1.299 0.168 -1.009 -0.338 -4.014 21.5 0.000
Headway to
KNUST
-0.409 1.519 0.196 -0.802 -0.017 -2.089 0.0029 0.041
Source: from study
Paired t-Test results in Table 7 showed there was no
significant (p> 0.05) saturation flow variation of
vehicular movements between the field and the
simulated situation for approaches from both direction.
Eta squared statistic 0.0028 indicated a small size effect;
meaning only 0.28% of the variation in the simulated
saturation flow was explained by the field saturation
flow from KNUST. Again Eta squared statistic 0.00042
indicated a small size effect which meant that only
4.2% of the variation in the simulated saturation flows
was explained by the field saturation to KNUST.
Paired t-Test results in Table 7 showed there was
significant (p<0.05) speed variation of vehicular
movements between the field and the simulated
situation for approaches from both direction. Eta
squared statistic 0.825 indicated a large size effect;
meaning 82.5% of the variation in the simulated speed
was explained by the field speed from KNUST. Again
Eta squared statistic 0.403 indicated a large size effect
which meant that 40.3% of the variation in the
simulated speeds was explained by the field speeds to
KNUST.
Paired t-Test results in Table 7 showed there was
significant (p<0.05) headway variation of vehicular
movements between the field and the simulated
situation for approaches from both direction. Eta
squared statistic 0.215 indicated a small size effect;
meaning 21.5% of the variation in the simulated
headway was explained by the field headway from
KNUST. Again Eta squared statistic 0.000029 indicated
a small size effect which meant that 0.0029% of the
variation in the simulated headways was explained by
the field headways to KNUST.
E. K. Nyantakyi, C. A. Adams, J. K. Borkloe and D. Pobee / International Journal of Engineering
Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 4, Jul-Aug 2013, pp.2566-2590
2578 | P a g e
Table 8: Results of paired Sample Test at Stadium Junction Intersection (Sat flow SIM)
Approaches Paired Difference t Eta
squared
(%)
Sig.
(2-
tailed)
p-value
Mean Standard
deviation
Standard
error
mean
95% confidence
interval of the
difference
Lower Upper
Saturation
from KNUST
-41.067 571.345 104.313 -254.410 172.277 -0.394 0.53 0.697
Saturation to
KNUST
0.633 509.752 93.068 -189.71 190.978 0.007 0.00017 0.995
Speed from
KNUST
1.900 2.280 0.416 1.049 2.751 4.565 41.8 0.000
Speed to
KNUST
2.800 0.997 0.182 2.428 3.172 15.389 89.1 0.000
Headway from
KNUST
0.5213 1.997 0.365 -0.225 1.267 1.430 6.59 0.164
Headway to
KNUST
0.623 0.913 0.167 -0.964 -0.283 -3.741 32.23 0.001
Source: from study
Paired t-Test results in Table 8 showed there was no
significant (p> 0.05) saturation flow variation of
vehicular movements between the field and the
simulated situation for approaches from both direction.
Eta squared statistic 0.0053 indicated a small size effect;
meaning only 0.53% of the variation in the simulated
saturation flow was explained by the field saturation
flow from KNUST. Again Eta squared statistic 61069.1 indicated a small size effect which meant
that only 0.00017% of the variation in the simulated
saturation flows was explained by the field saturation to
KNUST.
Paired t-Test results in Table 8 showed there was
significant (p<0.05) speed variation of vehicular
movements between the field and the simulated
situation for approaches from both direction. Eta
squared statistic 0.418 indicated a large size effect;
meaning 41.8% of the variation in the simulated speed
was explained by the field speed from KNUST. Again
Eta squared statistic 0.891 indicated a large size effect
which meant that 89.1% of the variation in the
simulated speeds was explained by the field speeds to
KNUST.
Paired t-Test results in Table 8 showed there was no
significant (p> 0.05) headway variation of vehicular
movements between the field and the simulated
situation from KNUST and significant (p<0.05)
headway variation of vehicular movements between the
field and the simulated situation to KNUST. Eta
squared statistic 0.0659 indicated a moderate size effect;
meaning 6.59% of the variation in the simulated
headway was explained by the field headway from
KNUST. Again Eta squared statistic 0.3223 indicated a
large size effect which meant that 32.23% of the
variation in the simulated headways was explained by
the field headways to KNUST.
Table 9: Results of paired Sample Test at Amakom
Intersection (Sat flow SIM)
Approaches
Paired Difference
t
Eta
squared
(%)
Sig.
(2-
tailed)
p-value
Mean Standard
deviation
Standard
error
mean
95% confidence
interval of the
difference
Lower Upper
Saturation
from KNUST
-45.800 728.645 94.068 -234.029 142.429 -0.487 0.4 0.628
Saturation to
KNUST
-2.683 580.118 74.893 -152.54 147.177 -0.036 0.0022 0.972
Speed from
KNUST
5.300 1.843 0.291 4.711 5.889 18.193 84.9 0.000
Speed to
KNUST
3.475 1.198 0.189 3.092 3.858 18.345 85.1 0.000
E. K. Nyantakyi, C. A. Adams, J. K. Borkloe and D. Pobee / International Journal of Engineering
Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 4, Jul-Aug 2013, pp.2566-2590
2579 | P a g e
Headway from
KNUST
-0.353 1.414 0.183 -0.719 0.119 -1.936 6.0 0.058
Headway to
KNUST
0.600 2.211 0.285 0.029 1.171 2.103 7.0 0.040
Source: from study
Paired t-Test results in Table 9 showed there was no
significant (p> 0.05) saturation flow variation of
vehicular movements between the field and the
simulated situation for approaches from both direction.
Eta squared statistic 0.004 indicated a small size effect;
meaning only 0.4% of the variation in the simulated
saturation flow was explained by the field saturation
flow from KNUST. Again Eta squared statistic
0.000022 indicated a small size effect which meant that
only 0.0022% of the variation in the simulated
saturation flows was explained by the field saturation to
KNUST. There was significant (p<0.05) speed
variation of vehicular movements between the field and
the simulated situation for approaches from both
direction. Eta squared statistic 0.849 indicated a large
size effect; meaning 84.9% of the variation in the
simulated speed was explained by the field speed from
KNUST. Again Eta squared statistic 0.851 indicated a
large size effect which meant that 85.1% of the
variation in the simulated speeds was explained by the
field speeds to KNUST. There was no significant
(p>0.05) headway variation of vehicular movements
between the field and the simulated situation from
KNUST and significant (p<0.05) headway variation of
vehicular movements between the field and the
simulated situation to KNUST. Eta squared statistic
0.060 indicated a small size effect; meaning 6.0% of the
variation in the simulated headway was explained by
the field headway from KNUST. Again Eta squared
statistic 0.070 indicated a moderate size effect which
meant that 7.0% of the variation in the simulated
headways was explained by the field headways to
KNUST.
3.2.3 Regression Analysis
In order to calibrate the model, a regression
analysis was carried out from and to KNUST
approaches at the selected intersections to establish
which of either speed or headway was a better predictor
or has a strong correlation with saturation flow. Table
10 shows the comparison of field and simulated
saturation flow, headway and speed for the selected
intersections.
Table 10: Comparison of Field and Simulated Saturation flow, Speed and Headway for Selected Intersections
Intersection Approach Conditions R R square
(R2)
Adjusted R
Square (R2)
Standard Error of
the Estimate
From KNUST Field 0.797 0.635 0.622 451.078
Anloga Simulated 1.000 1.000 1.000 0.000
To KNUST Field 0.826 0.683 0.672 408.457
Simulated 1.000 1.000 1.000 0.000
From KNUST Field 0.856 0.733 0.713 305.676
Stadium Simulated 1.000 1.000 1.000 0.000
To KNUST Field 0.922 0.850 0.839 205.364
Simulated 1.000 1.000 1.000 0.000
From KNUST Field 0.865 0.748 0.735 367.072
Amakom Simulated 1.000 1.000 1.000 0.000
To KNUST Field 0.827 0.685 0.668 337.738
Simulated 1.000 1.000 1.000 0.232
Source: from study
For Anloga, it was established that headway
had a strong correlation with saturation flow from
KNUST approach with R2
field = 0.635 and R2simuated =
1.000. Similarly headway had a strong correlation with
saturation flow to KNUST approach with R2field = 0.683
and R2simuated = 1.000
The adjusted R2 value was expressed in percentage as
62.2%. Thus the model (headway and speed) explained
62.2% of the variance in the saturation flows for field
condition as shown in Table 10..
The adjusted R2 value was expressed in percentage as
67.2%. Thus the model (headway and speed) explained
67.2% of the variance in the saturation flows for field
condition.
For Stadium, it was established that headway had a
strong correlation with saturation flow from KNUST
approach with R2field = 0.733 and R
2simuated = 1.000.
Similarly headway had a strong correlation with
E. K. Nyantakyi, C. A. Adams, J. K. Borkloe and D. Pobee / International Journal of Engineering
Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 4, Jul-Aug 2013, pp.2566-2590
2580 | P a g e
saturation flow to KNUST approach with R2
field = 0.85
and R2simuated = 1.000 as shown in Table 10.
The adjusted R2 value was expressed in percentage as
71.3%. Thus the model (headway and speed) explained
62.2% of the variance in the saturation flows for field
condition as shown in Table 10
The adjusted R2 value was expressed in percentage as
83.9%. Thus the model (headway and speed) explained
83.9% of the variance in the saturation flows for the
field condition.
For Amakom intersection, it was established
that headway had a strong correlation with saturation
flow from KNUST approach with R2field = 0.748 and
R2simuated = 1.000. Similarly headway had a strong
correlation with saturation flow to KNUST approach
with R2field = 0.683 and R
2simuated = 1.000 as can be seen
in Table 10.
The adjusted R2 value was expressed in
percentage as 73.5%. Thus the model (headway and
speed) explained 73.5% of the variance in the saturation
flows for field condition as shown in Table 10. The
adjusted R2 value was expressed in percentage as
66.8%. Thus the model (headway and speed) explained
66.8% of the variance in the saturation flows for field
condition.
The adjusted R2 value was expressed in
percentage as 100% for the selected intersections. This
therefore meant that the model (headway and speed)
explained 100% of the variance in the saturation flows
for simulated conditions for the selected intersections.
Table 11: Results of Field and Simulated Conditions co-efficients (From KNUST) Anloga Intersection
Predictors
Unstandardized
Coefficients
Standardized
Coefficients
t Sig. Collinearity Statistics
B Std.
Error
Beta Tolerance VIF
Constant 3362.416 576.527 5.832 0.000
SpeedFIELD -18.644 21.366 -0.070 -0.873 0.387 0.995 1.005
Headway for
KNUST
Approach
-463.630
47.144
-0.789
-9.834
0.000
0.995
1.005
Constant 1373.000 0.000 6 x 108 0.000
Speed for
KNUST
Approach
1.98x1015
0.000
0.000
0.000
1.000
0.738
1.355
HeadwaySIM 100.000 0.000 1.000 3 x 108 0.000 0.738 1.355
Source: from study
For the field condition, the model showed that headway
(-0.789) made a unique contribution in explaining the
saturation flow when the variance in the model was
controlled for as compared to that of speed (-0.070)
which made a small contribution to the model as can be
seen in Table 11. Headway had a significance level of
0.000 which was less than 0.05; thus contributed greatly
to the prediction of saturation flow whiles Speed (0.387)
made an insignificant contribution.
The field equation connecting saturation flow (q), speed
(u) and headway (h) is:
416.3362630.463644.18 huq Eq. (5)
For the simulated condition, the model showed that
headway (1.000) made the strongest contribution in
explaining the saturation flow when the variance in the
model was controlled for as compared to that of speed
(0.000) which made no contribution to the model as can
be seen in Table 11. Headway had a significance level
of 0.000 which was less than 0.05; thus contributed to
the prediction of saturation flow whiles Speed (1.000)
made an insignificant contribution.
The simulated equation connecting saturation flow (q),
headway (h) and speed (u) is:
1373100151098.1
huq Eq. (6)
Table 12: Results of Field and Simulated Conditions co-efficients (To KNUST) Anloga Intersection
Predictors
Unstandardized
Coefficients
Standardized
Coefficients
t Sig. Collinearity Statistics
B Std.
Error
Beta Tolerance VIF
Constant 3252.782 279.859 11.623 0.000
Speed to KNUST
Approach
-17.815
8.758
-0.153
-2.034
0.047
0.987
1.014
E. K. Nyantakyi, C. A. Adams, J. K. Borkloe and D. Pobee / International Journal of Engineering
Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 4, Jul-Aug 2013, pp.2566-2590
2581 | P a g e
Headway to
KNUST
Approach
-392.412
35.508
-0.830
-11.051
0.000
0.987
1.014
Constant 1459.000 0.000 9x108 0.000
SpeedSIM 3.94x10-14
0.000 0.000 0.000 1.000 0.767 1.304
HeadwaySIM 100.000 0.000 1.000 4x108 0.000 0.767 1.304
Source: from study
For the field condition, the model showed that
headway (-0.830) made a unique contribution in
explaining the saturation flow when the variance in the
model was controlled for as compared to that of speed
(-0.153) which made less contribution to the model as
shown in Table 12. Headway had a significance level of
0.000 which was less than 0.05, thus greatly contributed
to the prediction of saturation flow though speed (0.047)
made insignificant contribution.
The field equation connecting saturation flow (q), speed
(u) and headway (h) is:
782.3252412.392815.17 huq Eq. (7)
For the simulated condition, the model showed that
headway (1.000) made a unique contribution in
explaining the saturation flow when the variance in the
model was controlled for as compared to that of speed
(0.000) which made no contribution to the model as
shown in Table 12. Headway had a significance level of
0.000 which was less than 0.05, thus greatly contributed
to the prediction of saturation flow whiles speed (1.000)
made insignificant contribution.
The simulated equation connecting saturation flow (q),
speed (u) and headway (h) is:
1459100141094.3
huq Eq. (8)
Table 13: Results of Field and Simulated Conditions co-efficients (From KNUST) Stadium Intersection
Performance
Measures
Unstandardized
Coefficients
Standardized
Coefficients
t Sig. Collinearity Statistics
B Std.
Error
Beta Tolerance VIF
Constant 2404.311 508.995 4.724 0.000
Speed for
KNUST
Approach
-10.446 19.046 -0.055 -0.548 0.588 0.973 1.028
Headway for
KNUST
Approach
-242.446 28.918 -0.845 -8.384 0.000 0.973 1.028
Constant 949.000 0.000
SpeedSIM 4.3x10-14
0.000 0.000 0.756 1.323
HeadwaySIM 100.000 0.000 1.000 0.756 1.323
Source: from study
For the field condition, the model showed that
headway (-0.845) made a strong contribution in
explaining the saturation flow when the variance in the
model was controlled for as compared to that of speed
(-0.055) which made less contribution to the model as
can be seen in Table 13. Headway had a significance
level of 0.000 which was less than 0.05; thus greatly
contributed to the prediction of saturation flow whiles
speed (0.548) made insignificant contribution.
The field equation connecting saturation flow (q), speed
(u) and headway (h) is:
311.2404446.242446.10 huq Eq. (9)
For the simulated condition, the model showed that
headway (1.000) made a unique contribution in
explaining the saturation flow when the variance in the
model was controlled for as compared to that of speed
(0.000) which made no contribution to the model as can
be seen in Table 13.
The simulated equation connecting saturation flow (q),
speed (u) and headway (h) is:
94910014103.4
huq Eq. (10)
E. K. Nyantakyi, C. A. Adams, J. K. Borkloe and D. Pobee / International Journal of Engineering
Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 4, Jul-Aug 2013, pp.2566-2590
2582 | P a g e
Table 14: Results of Field and Simulated Conditions co-efficients (To KNUST) Stadium Intersection
Performance
Measures
Unstandardized
Coefficients
Standardized
Coefficients
t Sig. Collinearity Statistics
B Std.
Error
Beta Tolerance VIF
Constant 3057.240 571.109 5.353 0.000
SpeedField -5.109 23.856 -0.017 -0.214 0.832 0.917 1.090
Headway -528.143 44.824 -0.917 -11.783 0.000 0.917 1.090
Constant 1337.000 0.000
SpeedSIM 1.47x10-14
0.000 0.000 0.756 1.323
HeadwaySIM 100.000 0.000 1.000 0.756 1.323
Source: from study
For the field condition, the model established that
headway (0.917) made a unique contribution in
explaining the saturation flow when the variance in the
model was controlled for as compared to that of speed
(0.017) which made less contribution to the model as
shown in Table 14. Headway had a significance level of
0.000 which was less than 0.05, thus greatly contributed
to the prediction of saturation flow though speed (0.832)
made insignificant contribution.
The field equation connecting saturation flow (q), speed
(u) and headway (h) is:
240.3057143.528109.5 huq Eq. (11)
For the simulated condition, the model established that
headway (1.000) made a unique contribution in
explaining the saturation flow when the variance in the
model was controlled for as compared to that of speed
(0.000) which made no contribution to the model as
shown in Table 14.
The simulated equation connecting saturation flow (q),
speed (u) and headway (h) is:
1337100141047.1
huq Eq. (12)
Table 15: Results of Field and Simulated Conditions co-efficients (From KNUST) Amakom Intersection
Predictors
Unstandardized
Coefficients
Standardized
Coefficients
t Sig. Collinearity Statistics
B Std.
Error
Beta Tolerance VIF
Constant 2714.454 622.587 4.360 0.000
Speed from
KNUST
Approach
6.586
22.765
0.024
0.289
0.774
0.998
1.002
Headway from
KNUST
Approach
-476.444
45.546
-0.864
-10.461
0.000
0.998
1.002
Constant 1199.000 0.000 5x108 0.000
SpeedSIM 1.27x10-14
0.000 0.000 0.000 1.000 0.877 1.141
HeadwaySIM 100.000 0.000 1.000 3x108 0.000 0.877 1.141
Source: from study
For the field condition, the model established
that headway (-0.864) made a unique contribution in
explaining the saturation flow when the variance in the
model was controlled for as compared to that of speed
(0.024) which made a less contribution to the model as
can be seen in Table 15. Headway has a significance
level of 0.000 which was less than 0.05; thus greatly
contributed to the prediction of saturation flow whiles
speed (0.774) made insignificant contribution.
The field equation connecting saturation flow (q), speed
(u) and headway (h) is:
454.274144.476586.6 huq Eq. (13)
For the simulated condition, the model established that
headway (1.000) made the strongest contribution in
explaining the saturation flow when the variance in the
model was controlled for as compared to that of speed
(0.000) which made no contribution to the model as can
be seen in Table 15. Headway has a significance level
of 0.000 which made significant contribution to the
prediction of saturation flow whiles Speed (1.000)
made an insignificant contribution.
The simulated equation connecting saturation
flow (q), speed (u) and headway (h) is:
E. K. Nyantakyi, C. A. Adams, J. K. Borkloe and D. Pobee / International Journal of Engineering
Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 4, Jul-Aug 2013, pp.2566-2590
2583 | P a g e
1199100141027.1
huq Eq.
(14)
Table 16: Results of Field and Simulated Conditions co-efficients (To KNUST) Amakom Intersection
Predictors
Unstandardized
Coefficients
Standardized
Coefficients
t Sig. Collinearity
Statistics
B Std. Error Beta Tolerance VIF
Constant 3073.799 1014.829 3.029 0.004
Speed to
KNUST
Approach
-24.841
34.732
-0.083
-0.715
0.479
0.630
1.586
Headway to
KNUST
Approach
-333.976
44.364
-0.875
-7.528
0.000
0.630
1.586
Constant 983.996 1.376 715.028 0.000
Speed SIM 0.270 0.045 0.012 5.969 0.000 0.877 1.141
Headway SIM 101.318 0.203 1.004 499.941 0.000 0.877 1.141
Source: from study
For the field condition, the model established that
headway (-0.875) made the strongest contribution in
explaining the saturation flow when the variance in the
model was controlled for as compared to that of speed
(0.083) which made a less contribution to the model as
shown in Table 16. Headway had a significance level of
0.000 which was less than 0.05; thus greatly contributed
to the prediction of saturation flow whiles Speed (0.479)
made an insignificant contribution.
The field equation connecting saturation flow (q), speed
(u) and headway (h) is:
799.3073976.333841.24 huq Eq. (15)
For the simulated condition, the model established that
headway (1.004) made the strongest contribution in
explaining the saturation flow when the variance in the
model was controlled for as compared to speed (0.012)
which made a less contribution to the model as shown
in Table 16. Headway had a significance level of 0.000
which was less than 0.05; thus greatly contributed to the
prediction of saturation flow as well as Speed (0.000).
The simulated equation connecting saturation flow (q),
speed (u) and headway (h) is:
996.983318.10127.0 huq Eq. (16)
3.3 Comparison of Computed and Simulated
Performance Indicators
Computed manual performance measures at
the selected intersections were compared with the
simulated performance indicators generated by the
calibrated Synchro model as can be seen in Table 17.
Table 17: Comparison of Computed and Simulated Performance Indicators
Performance
Indicators
Anloga Stadium Amakom
Computed Adjusted Computed Adjusted Computed Adjusted
Cycle Length 210 200 68 65 172 193
v/c ratio 1.46 2.79 1.53 0.78 1.78 1.48
Int. Delay (s) 94 183.5 37 47.7 126 157.1
Int. LOS F F D D F F
ICU (%) 70.4 63.2 67.6
Offsets 135.0 41 133
Source: from study
From Table 17, Anloga and Amakom
intersections had level of service F. The level of service
F described a forced-flow operation at low speeds,
where volumes were below capacity. These conditions
usually resulted from queues of vehicles backing up a
restriction downstream. Speeds were reduced
substantially and stoppages occurred for short or long
periods of time because of the downstream congestion.
It represented worst conditions. Stadium intersection
had level of service D which approached unstable flow,
with tolerable operating speeds being maintained,
though considerably affected by changes in operating
conditions. Drivers had little freedom to maneuver, and
comfort and convenience were low. These conditions
could be tolerated, however, for short periods of time. It
represented intermediate conditions.
E. K. Nyantakyi, C. A. Adams, J. K. Borkloe and D. Pobee / International Journal of Engineering
Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 4, Jul-Aug 2013, pp.2566-2590
2584 | P a g e
3.4 Analysis of Alternative Phasing Plans 3.4.1 Change in phasing plan without Geometric
Improvement
Figs. 5 to 7 show the existing traffic situations together with their possible alternative phasing plans at the selected
intersections.
Anloga Intersection
Figure 5: Existing traffic with alternative phasing plans at Anloga intersection
Source: from study
Table 18: Results of Performance Indicators for three Different Alternatives
Performance
Indicators
Existing
Situation
Alternative 1 Alternative 2 Alternative 3
Cycle Length(s) 184 176 176 176
v/c ratio 2.77 1.25 2.65 2.65
Int. Delay (s) 183.5 103.2 170.7 170.7
Int. LOS F F F F
ICU (%) 70.4 70.4 70.4 70.4
Offsets 135.0 24 24 24
Source: from study
Out of the three possible alternative phasing plans at the
Anloga intersection, alternative 1 gave out the best
optimized cycle length of 176secs with an offset of
24secs and an intersection delay of 103.2secs as shown
in Table 18. These indicators compared to the existing
indicators were better in terms of cycle length, v/c ratio
and delay. Level of service F described a forced-flow
operation at low speeds, where volumes were below
capacity. These conditions usually resulted from queues
of vehicles backing up a restriction downstream. Speeds
were reduced substantially and stoppages occurred for
short or long periods of time because of the
downstream congestion. It represented worst conditions.
Stadium Intersection
E. K. Nyantakyi, C. A. Adams, J. K. Borkloe and D. Pobee / International Journal of Engineering
Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 4, Jul-Aug 2013, pp.2566-2590
2585 | P a g e
Figure 6: Existing traffic with alternative phasing plans at Stadium Junction intersection
Source: from study
Table 19: Results of performance indicators for three Different Alternatives
Performance
Indicators
Existing
Situation
Alternative 1 Alternative 2 Alternative 3
Cycle Length 65 65 60 60
v/c ratio 0.78 0.78 0.78 0.78
Int. Delay (s) 47.7 47.7 45.7 45.7
Int. LOS D D D D
ICU (%) 63.2 63.2 63.2 63.2
Offsets 41 41 41 41
Source: from study
Out of the three possible alternative phasing plans at the
Stadium Junction intersection, alternative 2 gave out the
best optimized cycle length of 60secs with an offset of
41secs and an intersection delay of 45.7secs as in Table
19. These indicators compared to the existing indicators
were better in terms of cycle length, v/c ratio and delay
and yet the LOS was still D. Level of service D
approached unstable flow, with tolerable operating
speeds being maintained, though considerably affected
by changes in operating conditions. Drivers had little
freedom to maneuver, and comfort and convenience
were low. These conditions could be tolerated, however,
for short periods of time. It represented intermediate
conditions. Site observations at the intersection
however showed that the traffic congestion was mostly
associated with the left turning traffic towards stadium.
The storage length for the vehicles was exceeded and
therefore the vehicles spilled back into the double lane
for the through traffic. This therefore reduced the
saturation flow for the through traffic at the intersection
when the indication turned green.
Amakom Intersection
E. K. Nyantakyi, C. A. Adams, J. K. Borkloe and D. Pobee / International Journal of Engineering
Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 4, Jul-Aug 2013, pp.2566-2590
2586 | P a g e
Figure 7: Existing traffic with alternative phasing plans at Amakom intersection
Source: from study
Table 20: Results of performance indicators for two Different Alternatives
Performance
Indicators
Existing
Situation
Alternative 1 Alternative 2
Cycle Length 193 170 172
v/c ratio 1.48 1.48 1.48
Int. Delay (s) 157.9 154.6 160.3
Int. LOS F F F
ICU (%) 67.6 67.6 67.6
Offsets 133 133 133
Source: from study
Out of the two possible alternative phasing
plans at the Amakom intersection, alternative 1 gave
out the best optimized cycle length of 170secs with an
offset of 133secs and an intersection delay of 154.6secs
as shown in Table 20. These indicators compared to the
existing indicators were better in terms of cycle length,
v/c ratio and delay. Level of service F described a
forced-flow operation at low speeds, where volumes
were below capacity. These conditions usually resulted
from queues of vehicles backing up a restriction
downstream. Speeds were reduced substantially and
stoppages occurred for short or long periods of time
because of the downstream congestion. It represented
worst conditions.
3.4.2 Change in phasing plan with Geometric
Improvement
The best alternative from the phasing plan in
Figures 5, 6 and 7 were further investigated upon by
adding lane(s) to the through put traffic at the selected
intersections.
Anloga Intersection
The best alternative from the phasing plan in
Fig. 5 was further investigated upon by adding lane(s)
to the through put traffic at Anloga intersection. Table
21 shows the summary of performance indicators with
geometric improvement.
Table 21: Summary of performance indicators with geometric improvement at Anloga Intersection
Performance
Indicators
Existing
Situation
Alternative 1 Addition of 1
lane
Addition of 2
lanes
Cycle Length 184 176 176 176
v/c ratio 2.77 1.25 0.86 0.78
Int. Delay (s) 183.5 103.2 62.7 58.6
Int. LOS F F E E
ICU (%) 70.4 70.4 62.0 57.1
Offsets 135.0 24 24 24
E. K. Nyantakyi, C. A. Adams, J. K. Borkloe and D. Pobee / International Journal of Engineering
Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 4, Jul-Aug 2013, pp.2566-2590
2587 | P a g e
Source: from study
There was the need to introduce geometric
improvement as correctional measure. With the
addition of one lane on the major approach through put
traffic, the v/c ratio reduced from 1.25 to 0.86 with a
corresponding reduction in the delay in Table 21. The
level of service also changed from F to E meaning that
there had been an improvement at the intersection.
When 2 lanes each were added to each major through
put approach, the v/c ratio again reduced from 0.86 to
0.78 with a corresponding reduction in the delay. The
level of service E could not be described by speed alone,
but represented operations at lower operating speeds,
typically, but not always, in the neighborhood of
30miles per hour, with volumes at or near the capacity
of the highway. Flow was unstable, and there may be
stoppages of momentary duration. This level of service
was associated with operation of a facility at capacity
flows. It represented intermediate conditions and
performance had improved slightly at the intersection.
Similarly, the existing reservation was further checked
against the geometric improvement.
Stadium Intersection
The best alternative from the phasing plan in Fig. 6 was
further investigated upon by adding lane(s) to the
through put traffic at the intersection. Table 22 shows
the summary of performance indicators with geometric
improvement.
Table 22: Summary of performance indicators with geometric improvement at Stadium Intersection
Performance
Indicators
Existing
Situation
Alternative 2 Addition of 1
lane
Addition of 2
lanes
Cycle Length 65 60 60 60
v/c ratio 0.78 0.78 0.78 0.78
Int. Delay (s) 47.7 45.7 40.7 39.0
Int. LOS D D D D
ICU (%) 63.2 63.2 56.2 52.1
Offsets 41 41 41 41
Source: from study
There was the need to introduce geometric
improvement as correctional measure. When one lane
was added to the major through put traffic from each
approach, the v/c ratio was 0.78 with a corresponding
reduction in delay as in Table 22. When 2 lanes each
were added to each major through put approach, the v/c
ratio was again 0.78 with a corresponding reduction in
delay. Level of service D approached unstable flow,
with tolerable operating speeds being maintained,
though considerably affected by changes in operating
conditions. Drivers had little freedom to maneuver, and
comfort and convenience were low. These conditions
could be tolerated, however, for short periods of time. It
represented intermediate conditions. Similarly, the
existing reservation was further checked against the
geometric improvement.
Amakom Intersection
The best alternative in Fig. 7 was further
investigated upon by adding lane(s) to the through put
traffic at Amakom intersection. Table 23 shows the
summary of performance indicators with geometric
improvement.
Table 23: Performance indicators with geometric improvement at Amakom Intersection
Performance
Indicators
Existing
Situation
Alternative 1 Addition of 1
lane
Addition of 2
lanes
Cycle Length 193 170 170 170
v/c ratio 1.48 1.48 1.03 0.82
Int. Delay (s) 157.9 154.6 73.9 60.4
Int. LOS F F E E
ICU (%) 67.6 67.6 49.8 43.6
Offsets 133 133 133 133
Source: from study
Geometric improvements were done as a
correctional measure. With the addition of one lane on
the major approach through put traffic, the v/c ratio
reduced from 1.48 to 1.03 with a corresponding
reduction in delay as seen in Table 23. The level of
service also changed from F to E meaning that there
had been an improvement at the intersection. When 2
lanes each were added to each major through put
E. K. Nyantakyi, C. A. Adams, J. K. Borkloe and D. Pobee / International Journal of Engineering
Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 4, Jul-Aug 2013, pp.2566-2590
2588 | P a g e
approach, the v/c ratio again reduced from 1.03 to 0.82
with a corresponding reduction in the delay. The level
of service E could not be described by speed alone, but
represented operations at lower operating speeds,
typically, but not always, in the neighborhood of
30miles per hour, with volumes at or near the capacity
of the highway. Flow was unstable, and there could be
stoppages of momentary duration. This level of service
was associated with operation of a facility at capacity
flows. It represented intermediate conditions and
performance had improved slightly at the intersection.
Similarly, the existing reservation was further checked
against the geometric improvement.
3.4.3 Grade Separation Option
With the continuous addition of lanes to the
through put traffic on the major approaches at Anloga
intersection, a situation may arise where there will not
be enough space to contain the addition of lanes. Since
the intersections will still be operating at a level of
service E, there is therefore the need to provide a
facility that can accommodate the traffic congestion at
the intersection on a long term basis. This then calls for
a grade separation (interchange) at the intersection to
allow the continuous flow of traffic.
3.4.4 Sensitivity Analysis
This was carried out to verify in detail the
sensitivity of the obtained results to the variation of the
input parameters at the selected intersections.
When the mean input parameters values from Table 5
such as speed and headway were substituted in the
modeled equations or obtained results as shown in
Table 24, the following saturation flow values were
obtained for the field and simulated conditions from
and to KNUST approaches at the selected intersections.
Table 24: Sensitivity Analysis of Saturation Flows for the Selected Intersections
Intersectio
n
Conditi
ons
Approach Obtained Results Speed Headw
ay
Saturat
ion
Flow
Field From
KNUST 416.3362630.463644.18 huq
26.63 2.59 1665
Anloga To
KNUST 782.3252412.392815.17 huq 28.43 2.56 1742
Simulat
ed
From
KNUST 1373100151098.1
huq
26.63 2.59 1632
To
KNUST 1459100141094.3
huq
28.43 2.56 1715
Field From
KNUST 311.2404446.242446.10 huq 26.63 2.59 1498
Stadium To
KNUST 240.3057143.528109.5 huq 28.43 2.56 1560
Simulat
ed
From
KNUST 94910014103.4
huq
26.63 2.59 1208
To
KNUST 1337100141047.1
huq
28.43 2.56 1971
Field From
KNUST 454.274144.476586.6 huq 26.63 2.59 1683
Amakom To
KNUST 799.3073976.333841.24 huq 28.43 2.56 1513
Simulat
ed
From
KNUST 1199100141027.1
huq
26.63 2.59 1458
To
KNUST 996.983318.10127.0 huq 28.43 2.56 1251
Source: from study
From the sensitivity analysis, it was deduced
from Table 24 that there was generally no variation
between the saturation flow values of the field and
simulated conditions from and to KNUST approaches
at the selected intersections. For Anloga intersection,
98.7% indicated that there was no variation between the
E. K. Nyantakyi, C. A. Adams, J. K. Borkloe and D. Pobee / International Journal of Engineering
Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 4, Jul-Aug 2013, pp.2566-2590
2589 | P a g e
saturation flow values of the field and simulated
conditions from the KNUST approach with 1.3%
indicating that there was variation between the
saturation flow values of the field and simulated
conditions from KNUST approach. Similarly, 98.4%
indicated that there was variation between the
saturation flow values of the field and simulated
conditions to KNUST approach with 1.6% indicating
that there was variation between the saturation flow
values of the field and simulated conditions to KNUST
approach.
For Stadium intersection, 80.4% indicated that
there was no variation between the saturation flow
values of the field and simulated conditions from the
KNUST approach with 19.4% indicating that there was
slight variation between the saturation flow values of
the field and simulated conditions from KNUST
approach. Similarly, 79.1% indicated that there was no
variation between the saturation flow values of the field
and simulated conditions to KNUST approach with
20.9% indicating that there was slight variation between
the saturation flow values of the field and simulated
conditions to KNUST approach.
For Amakom intersection, 86.3% indicated
that there was no variation between the saturation flow
values of the field and simulated conditions from the
KNUST approach with 13.7% indicating that there was
slight variation between the saturation flow values of
the field and simulated conditions from KNUST
approach. Similarly, 82.7% indicated that there was no
variation between the saturation flow values of the field
and simulated conditions to KNUST approach with
17.3% indicating that there was slight variation between
the saturation flow values of the field and simulated
conditions to KNUST approach.
The variations in the saturation flows for field
and simulated conditions was attributed to the fact that
headway made the strongest contribution in explaining
the saturation flow when the variance in the model was
controlled for as compared to that of speed which made
no contribution to the model. The sensitivity analysis
further confirmed that the obtained results from the
simulation model were as good as the model replicated
the specific real world characteristics of interest.
Generally, the obtained results showed no variations to
the input parameters from the sensitivity analysis from
and to KNUST approaches and that the obtained results
could be validated at intersections with similar Traffic
volume characteristics, roadway geometry and signal
timing.
IV. CONCLUSION The Chi square test and t-test analyses
revealed that headway had a strong correlation with
saturation flow for both field and simulated conditions
at 5% significance level. Again changes in phasing plan
without geometric improvement at the selected
intersections did not improve upon the overall
intersection’s level of service. Furthermore changes in
phasing plan with geometric improvement enhanced the
intersection’s level of service. The existing controls
needed to be replaced with a more effective, efficient
and reliable control scheme to ensure smooth and safe
operations. Stadium and Amakom signalized
intersections should be coordinated to allow as many
vehicles as possible to traverse those intersections
without any delay. This will reduce travel time of
motorists and also reduce congestion to the barest
minimum. An interchange should be constructed at the
Anloga intersection to allow free movement of vehicles
thereby minimizing congestion and accident
occurrences.The obtained results showed no variations
to the input parameters from the sensitivity analysis
from and to KNUST approaches and that, the obtained
results could be validated at intersections with similar
Traffic volume characteristics, roadway geometry and
signal timing. Obtained results from micro simulation
models can help the traffic engineer to comprehend the
problems existing and design improvement plans at the
intersection to reduce frequent delays and queue
spillbacks which will consequently improve upon the
levels of service at the selected intersections. Obtained
results could be used in forecasting the future traffic
conditions based on the present. Further research needs
to be investigated on the effect of left turning traffic at
Stadium intersection.
V. ACKNOWLEDGEMENT The authors would like to acknowledge the
management of Kumasi Polytechnic, Kumasi headed by
the Rector Prof. N.N.N. Nsowah-Nuamah, for
providing financial assistance and also Department of
Urban Roads (DUR), Kumasi for giving information on
signalized intersections in the Kumasi Metropolis.
Several supports from staff of the Civil Engineering
Department, KNUST and Kumasi Polytechnic, Kumasi
are well appreciated.
REFERENCES [1]. Drew, D.R. (1968). Traffic flow theory and
control. New York: McGraw-Hill
[2]. Kallberg, H. (1971). Traffic simulation (in
Finnish). Licentiate thesis, Helsinki University
of Technology, Transportation Engineering.
Espoo.
[3]. BCEOM and ACON Report (2004).
Consultancy Services for Urban Transport
Planning and Traffic Management Studies for
E. K. Nyantakyi, C. A. Adams, J. K. Borkloe and D. Pobee / International Journal of Engineering
Research and Applications (IJERA) ISSN: 2248-9622 www.ijera.com
Vol. 3, Issue 4, Jul-Aug 2013, pp.2566-2590
2590 | P a g e
Kumasi and Tamale for DUR (Ministry of
Transportation, Ghana), Chapter 5, pp. 37-52
[4]. Carole Turley (2007). Calibration Procedure
for a Microscopic Traffic Simulation Model. A
Thesis submitted to the faculty of Brigham
Young University, pp 2-19
[5]. Gerlough, D., and Huber, M. (1975) Traffic
flow theory. A monograph. TRB Special
Report 165. Washington, D.C.
[6]. McDonald, M., Brackstone, M., and Sultan, B.
(1998) Instrumented vehicle studies of traffic
flow models. Proceedings of the Third
International Symposium on Highway
Capacity, Volume 2, Ryysgaard, R., ed., pp.
755-774. Copenhagen: Transportation
Research Board and Danish Road Directorate.
[7]. Luo, L.H., Liu, H., Li, P., Wang, H., 2010.
Model predictive control for adaptive cruise
control with multi-objectives: comfort, fuel-
economy, safety and car-following. Journal of
Zhejiang University-SCIENCE A (Applied
Physics and Engineering), 11(3):191-201.
[doi:10.1631/jzus.A0900 374]
[8]. Tordeux, A., Lassarre, S., Roussignol, M.,
2010. An adaptive time gap car following
model. Transportation Research Part B:
Methodological, 44(8-9):1115-1131. [doi:10.
1016/j.trb.2009.12.018]
[9]. Brackstone, M., McDonald, M., 1999. Car-
following: a historical review. Transportation
Research Part F, 2(4):181-196.
[10]. Gazis, D.C., Herman, R., Rothery, R.W., 1961.
Follow-the leader models of traffic flow.
Operations Research, 9(4):545-567.
[doi:10.1287/opre.9.4.545]
[11]. Newell, G.F., 1961. Nonlinear effects in the
dynamics of car following. Operations
Research, 9(2):209-229.
[12]. Gipps, P.G., 1981. A behavioral car-following
model for computer-simulation.
Transportation Research Part B:
Methodological, 15(2):105-111.
[doi:10.1016/0191-2615 (81)90037-0]
[13]. Wiedemann, R., 1974. Simulation des Straß
Enverkehrsflusses Schriftenreihe des Instituts
für Verkehrswesen der Universität Karlsruhe
(in German)
[14]. Kikuchi, C., Chakroborty, P., 1992. Car
following model based on a fuzzy inference
system. Transportation Research Record,
1365:82-91.
[15]. Wu, J.P., Brackstone, M., McDonald, M.,
2000. Fuzzy sets and systems for a motorway
microscopic simulation model. Fuzzy Sets and
Systems, 116(1):65-76. [doi:10.1016/ S0165-
0114(99)00038-X]
[16]. Gunay, B., 2007. Car following theory with
lateral discomfort. Transportation Research
Part B: Methodological, 41(7): 722-735.
[doi:10.1016/j.trb.2007.02.002]
[17]. Mathew, T.V., Radhakrishnan, P., 2010.
Calibration of micro simulation models for
nonlane-based heterogeneous traffic at
signalized intersections. Journal of Urban
Planning and Development, 136(1):59-66.
[doi:10.1061/ (ASCE)0733-
9488(2010)136:1(59)]
[18]. Jin, S., Wang, D.H., Tao, P.F., Li, P.F., 2010.
Non-lane-based full velocity difference car
following model. Physica A: Statistical
Mechanics and Its Applications,
389(21):4654- 4662.
[doi:10.1016/j.physa.2010.06.014]
[19]. Transportation Research Board (1993).
Highway Capacity Manual, Chapter 9
(Signalized Intersections), National Research
Council, Washington, DC.
[20]. Turner and Harahap (1993). Simplified
Saturation Flow Data Collection Method:
Overseas Centre, Transport Research
Laboratory, Crowthorne Berkshire RG456AU,
United Kingdom PA1292/93
[21]. Liu, P., J. Lu, J. Fan, J. Pernia, and G.
Sakolow (1999). Effect of U-Turns on
Capacity of Signalized Intersections, in
Transportation Research Record 1920, Journal
of the Transportation Research Board, 2005,
pp. 74-80"
[22]. Minh and Sano (2003). Traffic Policy
Evaluations Using Micro Traffic Simulation:
A Case Study of Thailand, Paper for
Presentation at the 2003 Meeting of the
Transportation Research Board, Washington,
D.C., USA.